What is on this page
Almost everything written about product configurators is published by companies selling configurator software, which makes it good on features and quiet on what you have to prepare before any of that software will work. We build the 3D side of these projects and do not sell a platform, so this page is written from the other end.
What a 3D product configurator is
Nvidia defines it as a visualisation tool that lets users customise products in real time. In practice: the shopper picks a fabric, a finish, a size or a module, and the picture in front of them changes immediately, because it is not a picture. It is a 3D model being rendered live in the browser.
The visible half: swatches, hotspots and a scene that redraws as the shopper chooses.
The arithmetic that rules out photography
The case for a configurator is rarely about it being nicer. It is about a number that stops being payable.
Threekit uses the Herman Miller Cosm chair as the example: a chair listing at over 1,700 dollars with 27 customisable elements, which comes to 4,320 possible configurations. Photographing each one is not expensive, it is impossible. And a catalogue that cannot show a combination is a catalogue that cannot sell it with any confidence.
This is why configurators appear in the categories they do: furniture, joinery, machinery, apparel, jewellery, vehicles. Not because those buyers are more visual, but because those products have combinatorial ranges that no photographic budget survives.
The four parts, and who supplies each
| Part | What it does | Where it comes from |
|---|---|---|
| The 3D assets | The model, split into the pieces that can change, with every material built | A CGI studio. This is our part |
| The rules engine | Which combinations are allowed, what each one costs, what it does to lead time | Configurator software, or your own developers |
| The real time viewer | Draws the model in the browser and reacts to every click | The platform |
| Commerce plumbing | Turns a chosen configuration into a line item, a price and a production order | Your ecommerce or ERP system |
Being clear about that first row is the point of this page. We are not a configurator vendor and we do not want to sell you one. What projects fail on, repeatedly, is that the platform gets chosen and budgeted while the assets are treated as an afterthought, and then nothing can be switched on because the model was never built to come apart.
What the 3D model has to be
Every element the shopper can change has to exist as its own object before any of it can be swapped.
A model built for images is one object with one set of materials. A model for a configurator is a different piece of work:
- Split along the choices. If the shopper can change the arm, the arm is a separate object. This sounds obvious and is the single most common reason a model has to be rebuilt.
- Named predictably. The rules engine addresses parts by name. Inconsistent naming across a range turns integration into manual work that repeats with every new product.
- Light enough to run live. This is a real time model, not a render model. For reference, Shopify accepts GLB files up to 500 MB but names around four megabytes as ideal and optimises anything over fifteen.
- Materials as a set, not as a look. Forty fabrics means forty material definitions that behave consistently under the same lighting, not forty separate renders.
- Dimensionally correct. The configurator often feeds the price and sometimes the production order. A model that is approximately right is worse than no model.
The file formats and the weight targets are the same ones AR needs, and they are set out on our AR ready 3D models page. A catalogue prepared properly for one is largely prepared for the other.
The different types of configurator
| Type | What it is for | What it demands |
|---|---|---|
| Visual product configurator | The shopper sees the choice. Consumer facing | Best looking assets, lightest files |
| CPQ, configure price quote | Sales teams building complex quotes | Rules and pricing accuracy over visual polish |
| Parametric or engineering | Dimensions drive the geometry, output feeds manufacturing | CAD accuracy, not appearance |
| Room or scene planner | Several products arranged together in a space | A whole library built to one consistent standard |
The distinction that matters commercially is the first two. A visual configurator is a shop window and is judged on whether it looks like the product. CPQ is an internal tool and is judged on whether the quote is right. Buying one when you needed the other is the expensive mistake in this category.
What it costs, split honestly
Two of the possible combinations. Both came off one model, which is the whole economic argument.
| Line | Who charges it | What drives it |
|---|---|---|
| Platform | The configurator vendor | Usually a subscription, by traffic or by number of products |
| Integration | Your developers or the vendor | How cleanly your catalogue data describes the options |
| 3D assets | A studio | Number of products, how many parts each splits into, number of materials |
| Maintenance | Mostly the studio | Every new product and every new fabric added later |
On our side the work is billed at 20 dollars an hour, which puts a typical product around 160 dollars as a single model and a complex one around 360 before it is split for configuration. Splitting and material sets are quoted on top, and the honest planning note is that the assets are a recurring line rather than a one off: every product you add later needs the same treatment. Full rates are on 3D product modeling services.
Vendors publish encouraging numbers, and they should be read as what they are. Threekit reports that its own clients average a 40 percent increase in conversion and 20 percent in average order value. That is a company selling configurators describing its own customers, which does not make it wrong, but it is not independent evidence either.
When you do not need one
- Under about a dozen combinations. Photograph or render them all and put them in the gallery. Cheaper, faster, and nothing to maintain.
- The choice does not change the look. If the options are warranty length or delivery slot, a configurator is a dropdown with a 3D bill attached.
- Nobody owns the rules. Somebody internally has to know which combinations are actually manufacturable. If that knowledge lives in one person’s head, the project stalls there rather than in the 3D.
- The catalogue data is a mess. A configurator makes an inconsistent option structure visible to customers. Fix the data first.
FAQ
If you are scoping a configurator, the useful first step is counting: how many products, how many changeable elements on each, how many materials across the range. Send us that count and we will tell you what the asset side costs, whichever platform you end up choosing.
Ian Diev
Content Writer, Editor
Ian loves writing about innovative 3D technologies and their impact on architecture and design. In his spare time, he makes indie music, watches obscure movies, and cooks culinary masterpieces for friends.